Why BI-RADS?: Overview of Breast Imaging Reporting and Data System (BI-RADS)

History of BI-RADS

The 1980 s brought a sharp increase in mammography. In the early 1980s, only 15% to 20% of women had ever undergone a mammogram, and by the dawn of the 1990 s, 65% had participated in mammography at least once. This increase was largely due to favorable data showing effectiveness of screening mammography to reduce breast cancer mortality in multiple randomized controlled trials coupled with contemporaneous improvements in preoperative needle localization, making it easier to obtain a tissue diagnosis for suspicious lesions identified at mammography. As mammography utilization grew in the United States during the 1980s, significant inconsistencies became apparent in mammography quality, radiation dose, radiologist interpretative skills, and result reporting. Due to the concerns raised by breast imaging specialists and the American Medical Association (AMA), the American College of Radiology (ACR) convened two committees: the voluntary Mammography Accreditation Program (MAP) in 1987 and the Breast Imaging Reporting and Data System (BI-RADS) committee in 1988. The BI-RADS committee was charged with developing guidelines for standardized mammography reporting and management recommendations.

Five years later, the ACR BI-RADS committee, composed of academic breast imagers and private practice radiologists and chaired by Dr. Carl J. D’Orsi, released the first edition of the BI-RADS lexicon in 1993. This early BI-RADS document was robust and consisted of recommendations for (1) performance of screening and diagnostic mammography, (2) structure of the mammography report, (3) introduction of a mammography lexicon, and (4) final assessment categories with their respective management recommendations. It should be noted that the authors fashioned BI-RADS with a clear understanding that it would be a malleable and adaptable reporting and data system with the ability to transform, improve, and expand as needed to incorporate the continuous new advances in breast imaging technology, research, and patient care.

After the successful release of the ACR MAP in 1987 and just before the 1993 ACR BI-RADS committee release of the first edition of BI-RADS, the 1992 Mammography Quality Standards Act (MQSA) became federal law and was enacted, making breast imaging one of the most federally regulated medical specialties in the United States. By 1999, MQSA rules mandated, among other requirements, that every mammographic report include text for the final assessment category similar to that in the then-current 1998 third edition of BI-RADS. This new post-MQSA era in breast imaging marked a transition from inconsistent imaging quality and reporting to uniform mammography examination quality, interpretation, and reporting standards. MQSA regulations and accreditation requirements are covered in Chapter 20 .

As envisioned, BI-RADS has undergone four revisions since its first edition in 1993 (1995, 1998, 2003, and 2013), with each carefully crafted to improve clarity, patient management, and quality assurance ( Fig. 2.1 ). In 1998, the third edition included the first BI-RADS atlas with first-rate illustrations depicting each lexicon descriptor.

The fourth edition introduced many new lexicon descriptors such as the original asymmetry family (asymmetry, focal asymmetry, global asymmetry), division of suspicious calcifications into “intermediate risk” and “higher probability of malignancy,” and the option to subclassify BI-RADS final assessment category 4 into subcategories (4 A, 4B, and 4 C) to better communicate the risk of malignancy to both the patient’s health care provider and the pathologist interpreting a woman’s biopsy tissue specimens. Furthermore, the fourth edition (2003) introduced the first BI-RADS ultrasound (US) and BI-RADS magnetic resonance imaging (MRI) lexicons. These new US and MRI lexicons were designed to mirror the same lexicon descriptors used in mammography whenever possible (e.g., mass, shape, and margin) but also allow for new modality-specific descriptors such as orientation and echo pattern in US or foci/focus and kinetic curve assessment in MRI. The current fifth edition (2013) further refines the lexicon using consistent terminology across mammography, US, and MRI sections with evidence-based justification and better quality images for all sections including mostly digital images for mammography. This edition also defines auditing rules that apply to all three sections and allows uncoupling of the final assessment category from its management recommendation as discussed in the section below.

The BI-RADS Lexicon

The BI-RADS lexicon is a carefully constructed dictionary of descriptive terms for breast imaging findings, which has been published in numerous languages. It allows all radiologists to speak the same language and generate standard reports that are easily understood by referring providers and ancillary medical staff. The over 700-page BI-RADS Atlas provides detailed, annotated example images that illustrate the proper use of the lexicon terminology. Pathology of described findings is provided whenever possible. In addition to mammography, the most recent edition of the BI-RADS Atlas (2013) also includes comprehensive lexicons for breast US and MRI, plus a supplement with specific guidance for digital breast tomosynthesis (DBT). Specifics of lexicon and lesion descriptors are described in the relevant modality-specific chapters to follow.

The BI-RADS lexicon is systematically organized in a uniform branching format for each imaging modality. The organization is first divided at the highest level by imaging modality (mammography, US, MRI) followed by breast tissue and findings. Breast tissue is classified by its composition on mammography and by the amount of fibroglandular tissue and background parenchymal enhancement (BPE) on MRI. Breast composition in mammography refers to a woman’s individual admixture of dense fibroglandular tissue and fat within her breast (i.e., her breast density) and ranges from almost entirely fatty to extremely dense ( Fig. 2.2 ). Similar terms are used to describe the amount of fibroglandular tissue on MRI. Sonographic tissue composition can also be described as homogeneous (fatty or fibroglandular) versus heterogeneous echotexture; however, these terms may only be used for screening and/or whole breast diagnostic US when the breast tissue can be assessed in its entirety. Further details on breast composition are outlined in the section titled “The Structured Breast Imaging Report” and subsequent chapters dedicated to specific imaging modalities.

Findings are divided by (1) type (i.e., mass), (2) features of each finding type, and (3) a list of descriptive terms. For example, a mass (a finding type) is characterized by three features: shape, margin, and density. Descriptive terms for the shape of a mass include oval, round, or irregular.

The lexicon has been developed such that selection of the correct descriptor leads to use of the appropriate final assessment category and management recommendation. This correlation applies across the full spectrum of breast imaging findings, ranging from completely benign to highly suspicious for breast cancer. If suspicious lexicon terminology is used, tissue diagnosis should be recommended. If benign lexicon terminology is used, routine annual follow-up should be recommended. For example, if calcifications are described as having a “fine-linear branching” morphology, they are suspicious and require a biopsy. If the calcifications are “popcorn-like,” they are typically benign and can be left alone ( Fig. 2.3 ). If calcifications are distributed in a “segmental” pattern, there is a high likelihood of breast cancer and tissue sampling is required. Combining a suspicious descriptor with a benign final assessment and routine annual follow-up recommendation is not advised. Improper use of BI-RADS causes confusion among other radiologists and referring health care providers and potentially leads to mismanagement of the patient.

Multiple lexicon descriptors may be used to describe a single breast finding, but the overall management assessment for that finding should match the most suspicious BI-RADS lexicon descriptor used. An example would be a mass seen on MRI that is circumscribed and T2-hyperintense but demonstrates a heterogeneous internal enhancement pattern. Circumscribed is a more benign descriptor, but heterogeneous internal enhancement is a suspicious descriptor, and therefore this mass should be biopsied ( Fig. 2.4 ).

The BI-RADS lexicon is largely a data-driven document. For example, amorphous calcifications are considered moderately suspicious (BI-RADS 4B; Table 2.1 ) because studies have shown a 20% likelihood of association with malignancy. In some instances where supporting data is lacking, guidance is derived from expert opinion by the BI-RADS committee in an effort to make the Atlas a practical and useful resource. Remember, BI-RADS is an ever-evolving document that is modified and updated as new knowledge is acquired.

Special Considerations for Other Imaging Modalities

Whenever possible the BI-RADS lexicon strives to use the same lexicon terminology across all three modalities to maximize consistency and facilitate learning. As such, some lexicon descriptors are universal for all modalities. For example, the shape of a mass is described by the descriptive terms round , oval , or irregular in all modalities. Likewise, the margin of a mass is either circumscribed or not circumscribed in all modalities. Some differences between the three modalities are inevitable; for example, the BI-RADS language used to describe the margins of noncircumscribed masses (i.e., angular is specific to US, obscured is specific to mammography). The words used to describe the distribution ( diffuse , regional , linear , segmental ) of a non-mass finding are similar whether referring to calcifications on mammography or non-mass enhancement on MRI, with some small exceptions (i.e., grouped for calcifications and focal for non-mass enhancement).

Finally, some lexicon features are unique to a modality. For example, the shape of individual calcifications is only appreciable on mammography, and therefore morphology of calcifications is only described mammographically. The orientation of a mass is reserved for US imaging and refers to the orientation of the long axis of the mass relative to the skin line ( parallel is more often benign, not parallel is suspicious; Fig. 2.5 ).

Specific guidance for DBT is included in a supplement to the mammography section of the ACR BI-RADS fifth edition. The mammography BI-RADS lexicon terminology is fully applicable to DBT. With DBT, multiple very low-dose projections are obtained in an arc, which allow reconstruction of those images into thin slices of breast tissue. These slices, when viewed like the pages of a book, sequentially remove overlying layers of fibroglandular breast tissue, occasionally exposing otherwise hidden or obscured breast cancers. Furthermore, many “asymmetries” on two-dimensional (2D) mammography are readily shown to be summation artifact (overlapping normal fibroglandular breast tissue at different depths within the breast) on DBT resulting in fewer patient recalls. In addition, the margins of masses are often better seen on DBT compared with 2D mammography, such that additional mammographic views may not always be necessary. In these situations, it is recommended that patients are still recalled—even if the mass is circumscribed—for additional evaluation with US to fully characterize the mass. DBT is covered in more detail in Chapter 6 .

BI-RADS Final Assessment Categories

After using appropriate lexicon descriptive terms to describe findings in the body of the report, the next step is to assign a BI-RADS assessment category for the examination. If desired, separate BI-RADS assessments may be assigned to left and right breasts. The BI-RADS categories range from 0 to 6 and are divided into incomplete (category 0) and final assessments (categories 1–6). Each category is associated with a standardized management recommendation (see Table 2.1 ). In most cases, the management recommendation should match the assessment category; however, the fifth edition of the BI-RADS Atlas allows for uncoupling of the category and management under certain specific circumstances. Examples include women with appropriately categorized, probably benign findings (category 3) who desire biopsy for confirmation or a large painful cyst appropriately categorized as category 2 (benign) for which US-guided aspiration is recommended for symptomatic relief.

There are two types of breast imaging exams: screening and diagnostic. Screening mammography is an examination of an asymptomatic woman designed to detect clinically occult breast cancer. Diagnostic mammography is an examination of a symptomatic patient such as a woman with a breast lump or for a patient with a recent abnormal screening mammogram. The BI-RADS assessment categories are applicable to both of these imaging examination types; however, some assessment categories are more likely to be used in the setting of screening (BI-RADS 0) and some should be used only after a full diagnostic evaluation (BI-RADS 3, 4, 5, and 6). It is important to remember that a negative imaging evaluation can never guarantee that a woman does not have breast cancer, so any clinically suspicious lump or other breast symptom not explained by the diagnostic imaging evaluation should be referred to a surgeon to determine whether surgical biopsy is indicated in light of the negative imaging.

Box 2.1 summarizes key points for each of the seven assessment categories and their respective management recommendations. When the BI-RADS lexicon is followed, each assessment category is associated with a specific, rigorously proven likelihood of malignancy. The success of the BI-RADS lexicon and its final assessment categories is evident by its international adoption as the breast imaging reporting standard.